How to Fetch Real-Time Spot Bitcoin & Ethereum ETF Flows in Cursor & Claude (Free MCP Server)

#crypto #python #ai #web3

Institutional flow data for Spot Bitcoin (BTC) and Ethereum (ETH) ETFs is one of the most critical market sentiment indicators in crypto today. However, querying this data programmatically often means dealing with expensive enterprise APIs or fragile scrapers blocked by Cloudflare anti-bot walls.

To solve this, I built ETF Flow — an open-source, high-precision API and Model Context Protocol (MCP) server designed specifically for AI agents (Cursor IDE, Claude Desktop, Windsurf) and algorithmic developers.

⚡ Why Use an MCP Server for ETF Flows?

Instead of context-switching to web browsers or parsing raw web tables, native Model Context Protocol (MCP) integration allows your AI agent (like Cursor or Claude) to fetch real-time net inflow/outflow metrics directly inside your conversation context.

You can ask your AI:

"What were the net Bitcoin ETF inflows over the past 5 trading days?"

"Compare yesterday's BTC ETF flows against ETH ETF flows."

And your AI agent executes the tool call dynamically!

🚀 Quick Setup in Cursor & Claude Desktop

1. Configure the MCP Server

Add the following block to your MCP configuration (claude_desktop_config.json or Cursor MCP settings):

{
  "mcpServers": {
    "etf-flows": {
      "command": "uv",
      "args": [
        "--directory",
        "/path/to/free-etf-flows-mcp",
        "run",
        "python",
        "mcp_server.py"
      ]
    }
  }
}

2. Querying via REST API

If you prefer standard HTTP GET requests (Python, cURL, Node.js):

curl -X GET "https://ubzimdhjaqeirdhhwzug.supabase.co/functions/v1/smooth-handler?ticker=BTC&limit=10"

Sample JSON Response:

{
  "success": true,
  "count": 10,
  "data": [
    { "flow_date": "2026-07-28", "ticker": "BTC", "net_flow_usd": 154200000.00 },
    { "flow_date": "2026-07-27", "ticker": "BTC", "net_flow_usd": -45100000.00 }
  ]
}